TH‐E‐17A‐04: Geometric Validation of K‐Space Self‐Gated 4D‐MRI Vs. 4D‐CT Using A Respiratory Motion Phantom

Abstract

Purpose:4D‐CT is often limited by motion artifacts, low temporal resolution, and poor phase‐based target definition. We recently developed a novel k‐space self‐gated 4D‐MRI technique with high spatial and temporal resolution. The goal here is to geometrically validate 4D‐MRI using a MRI‐CT compatible respiratory motion phantom and comparison to 4D‐CT.Methods:4D‐MRI was acquired using 3T spoiled gradient echo‐based 3D projection sequences. Respiratory phases were resolved using self‐gated k‐space lines as the motion surrogate. Images were reconstructed into 10 temporal bins with 1.56×1.56×1.56mm3. A MRI‐CT compatible phantom was designed with a 23mm diameter ball target filled with highconcentration gadolinium(Gd) gel embedded in a 35×40×63mm3 plastic box stabilized with low‐concentration Gd gel. The whole phantom was driven by an air pump. Human respiratory motion was mimicked using the controller from a commercial dynamic phantom (RSD). Four breathing settings (rates/depths: 10s/20mm, 6s/15mm, 4s/10mm, 3s/7mm) were scanned with 4D‐MRI and 4D‐CT (slice thickness 1.25mm). Motion ground‐truth was obtained from input signals and real‐time video recordings. Reconstructed images were imported into Eclipse(Varian) for target contouring. Volumes and target positions were compared with ground‐truth. Initial human study was investigated on a liver patient.Results:4D‐MRI and 4D‐CT scans for the different breathing cycles were reconstructed with 10 phases. Target volume in each phase was measured for both 4D‐CT and 4D‐MRI. Volume percentage difference for the 6.37ml target ranged from 6.67±5.33 to 11.63±5.57 for 4D‐CT and from 1.47±0.52 to 2.12±1.60 for 4D‐MRI. The Mann‐Whitney U‐test shows the 4D‐MRI is significantly superior to 4D‐CT (p=0.021) for phase‐based target definition. Centroid motion error ranges were 1.35–1.25mm (4D‐CT), and 0.31–0.12mm (4D‐MRI).Conclusion:The k‐space self‐gated 4D‐MRI we recently developed can accurately determine phase‐based target volume while avoiding typical motion artifacts found in 4D‐CT, and is being further studied for use in GI targeting and motion management.This work supported in part by grant 1R03CA173273‐01